JOURNAL OF PALLIATIVE MEDICINE Volume 17, Number 12, 2014 ª Mary Ann Liebert, Inc. DOI: 10.1089/jpm.2014.0184
Cost Savings and Enhanced Hospice Enrollment with a Home-Based Palliative Care Program Implemented as a Hospice–Private Payer Partnership Christopher W. Kerr, MD, PhD,1 Kathleen A. Donohue, MA,2 John C. Tangeman, MD, FACP,1 Amin M. Serehali, MA,2 Sarah M. Knodel, MS, MPH,2 Pei C. Grant, PhD,1 Debra L. Luczkiewicz, MD,1 Kathleen Mylotte, MD,2 and Melanie J. Marien, MS, PA-C1
Abstract
Background: In the United States, 5% of the population is responsible for nearly half of all health care expenditures, with a large concentration of spending driven by individuals with expensive chronic conditions in their last year of life. Outpatient palliative care under the Medicare Hospice Benefit excludes a large proportion of the chronically ill and there is widespread recognition that innovative strategies must be developed to meet the needs of the seriously ill while reducing costs. Objective: This study aimed to evaluate the impact of a home-based palliative care program, implemented through a hospice–private payer partnership, on health care costs and utilization. Methods: This was a prospective, observational database study where insurance enrollment and claims data were analyzed. The study population consisted of Home Connections (HC) program patients enrolled between January 1, 2010 and December 31, 2012 who subsequently expired (n = 149) and who were also Independent Health members. A control group (n = 537) was derived using propensity-score matching. The primary outcome variable was overall costs within the last year of life. Costs were also examined at six months, three months, one month, and two weeks. Inpatient, outpatient, ancillary, professional, and pharmacy costs were compared between the two groups. Medical service utilization and hospice enrollment and length of stay were also evaluated. Results: Cost savings were apparent in the last three months of life—$6,804 per member per month (PMPM) cost for palliative care participants versus $10,712 for usual care. During the last two weeks of life, total allowed PMPM was $6,674 versus $13,846 for usual care. Enhanced hospice entry (70% versus 25%) and longer length of stay in hospice (median 34 versus 9 days) were observed. Conclusions: Palliative care programs partnered with community hospice providers may achieve cost savings while helping provide care across the continuum. Introduction
O
utpatient palliative care in most communities has been administered primarily via the Hospice Medicare Benefit which provides care for terminally ill individuals with an estimated life expectancy of six months or less. In addition, patients under the Hospice Medicare Benefit have chosen to ‘‘forgo curative treatment,’’1 which may exclude many patients facing chronic, incurable illness who might otherwise benefit from palliative care. Health care systems and payers are seeking innovative programs to address the needs of the most seriously ill patients who incur the highest
costs; in 2009, the sickest 5% of patients in the United States accounted for greater than 50% of health care spending, with a large proportion spent in the last year of life, often on hospital and/or ICU care.2 Hospital-based palliative care consultation services have, over the past decade, been offered at an increasing number of large U.S. hospitals,3 with a solid evidence base supporting financial and clinical efficacy.4–9 As noted in a recent review, less is known about the impact of outpatient palliative care programs.10 Palliative care offers the rare alignment between positive clinical and economic outcomes. A previous study of homebased palliative care showed reduced costs of up to 33%;11
1
Center for Hospice and Palliative Care, Cheektowaga, New York. Independent Health, Buffalo, New York. Accepted August 14, 2014. This article shares copyright with Independent Health, Williamsville, New York.
2
1328
HOME-BASED PALLIATIVE CARE
however, this program was implemented through a staff model insurance company and may be difficult to replicate outside of this framework. Similarly, the Aetna Compassionate Care Program showed a 22% reduction in costs at end of life while doubling hospice utilization.12 The current study describes economic outcomes of a collaborative model between a notfor-profit hospice and a private not-for-profit commercial insurance payer to provide home-based palliative care upstream and outside of hospice. This innovative program, implemented in Erie County in western New York, was supported via a per member per month (PMPM) fee. In a previous study,13 the clinical impact of this program was evaluated and was shown to improve symptom management in multiple symptom domains. More than two thirds of enrollees completed actionable advance directives, and the site of death was home for nearly half of those who expired during or after participation. Additionally, patients, caregivers, and physicians all gave the program high satisfaction scores.13 An economic analysis of participants in the program who expired is now reported, including the impact of this model on overall end-of-life costs, service-specific costs, and medical service utilization. Hospice enrollment and length of stay are also discussed. Methods Program description
The Center for Hospice & Palliative Care, Cheektowaga, New York, established the Home Connections (HC) homebased palliative care program in 2008 jointly with Independent Health, a commercial insurance provider. HC is intended for adults with advanced chronic illness upstream of hospice, so patients may still be receiving aggressive or curefocused treatments. The program is not restricted to patients with a prognosis of six months or less. The HC team includes a palliative care-trained registered nurse (RN), social worker, volunteers, and palliative care physician. Services include symptom management, education, supportive discussions about heath care decision making and goals, social work visits to facilitate access to community support services, respite care, and 24/7 on-call palliative care nurse support. Nurses and social workers visit participants a minimum of every 30 days, and a biweekly interdisciplinary group meeting is held to discuss each patient case with the palliative care physician. When clinically appropriate, HC also helps patients transition into hospice care, if desired. Sample
This study was approved by the University at Buffalo Health Sciences IRB. The study population consisted of HC participants enrolled for greater than 30 days between January 1, 2010 and December 31, 2012 who subsequently expired (n = 149). A control group was derived using propensity-score matching, to compare members with similar clinical and demographic characteristics (n = 537). Members enrolled less than 30 days (n = 68) were excluded, as many of those members enrolled directly in hospice (63%), therefore limiting the ability to measure the effect of HC on cost and utilization over a defined time period. Data was abstracted from insurance enrollment and claims. Logistic regression analysis was used to identify control group patients for propensity matching using age, gender,
1329
insurance product, median household income (categorized), neighborhood percentage minority (categorized), prospective risk score,14 and diagnoses historically associated with palliative care,5,16 These diagnoses included cancer, asthma, coronary artery disease (CAD), chronic obstructive pulmonary disease (COPD), diabetes, congestive heart failure (CHF), cerebral vascular accident (CVA), Alzheimer’s disease, dementia, chronic renal disease (CRD), and depression. The predicted probability of HC membership was grouped into quintiles, following the methodology of R. D’Agostino.17 The SAS GMATCH algorithm18 matched HC members to controls, at a 1:4 ratio. Variables found to be significantly related to participation in HC included age; neighborhood median household income, neighborhood percent minority race; diagnoses of cancer, diabetes, Alzheimer’s disease, dementia, CRD, and depression; insurance product; and prospective risk score. Analysis was restricted to members who had expired in order to eliminate bias introduced by trying to ‘‘match’’ clinical characteristics based on dates of enrollment. The data were filtered to only include those in the range of probability scores for HC members, yielding 5400 members in the pool used to match cases to controls. Statistical analyses
Outcome measures were costs and utilization of services, evaluated at two years, one year, six months, three months, one month, and two weeks prior to death. Testing of categorical factors was performed using the chi-square or Fisher’s exact test. Cost and risk score data were analyzed using the MannWhitney test. Statistical significance was set at 5% (two-sided). The effects of program membership on overall costs were analyzed using a repeated measures analysis of variance.19 This model allowed for changes in cost to be measured between the members based on program membership as well as over time. Cost data in the repeated measures analysis was transformed using a log function to adjust for skewing and outliers. The cost total within the last three months of life was modeled as a gamma distribution using a log link function20 using the generalized linear model procedure.21 Multivariate analysis adjusted for the effects of other demographic and clinical characteristics on overall costs. Results HC participants and comparison members
The median time in HC, excluding those enrolled < 30 days, was 118 days (range: 31–843). Demographic characteristics of HC participants and propensity-matched comparison members are shown in Table 1, with medical diagnoses in Table 2. Propensity score quintiles were evenly distributed, indicating that the groups were well matched for the likelihood of participation in HC. Demographic characteristics and diagnosis were similarly distributed between HC members and controls with the exception of CHF (40% of controls versus 28% of HC members [p = 0.01]) and CRD (28% of controls versus 14% of HC participants [p < 0.0001]). Aggregate costs
A comparison of overall total allowed costs over time is shown in Figure 1. Costs between the two groups were
1330
KERR ET AL.
Table 1. Distribution of Demographic Characteristics between Home Connections Participants and Propensity-Matched Comparison Members
Characteristic Propensity score quintile 1 2 3 4 5 Age group (years)a 0–50 51–65 66–75 76–90 ‡ 91 Neighborhood median household income ($)a £ 10,000 10,001–20,000 20,001–30,000 30,001–50,000 50,001–75,000 ‡ 75,001 Neighborhood percent minority racea < 50% ‡ 50% Insurance producta Article 43 Encompass Medicare Medisource Self-funded Gender Male Female Prospective risk scorea Median Range
Home connections (n = 149) No. (%)
Comparison members (n = 537) No. (%)
p-value 0.99
29 29 32 29 30
(19%) (19%) (21%) (19%) (20%)
107 113 108 104 105
(20%) (21%) (20%) (19%) (20%)
4 20 36 80 9
(3%) (13%) (24%) (53%) (6%)
9 71 116 280 61
(2%) (13%) (22%) (52%) (11%)
2 1 12 60 45 29
(1%) ( < 1%) (8%) (40%) (30%) (19%)
6 9 52 183 208 79
(1%) (2%) (10%) (34%) (39%) (15%)
0.37
0.26
0.60 136 (91%) 13 (9%)
498 (93%) 39 (7%)
5 10 131 3 0
15 27 470 19 6
0.52 (3%) (7%) (88%) (2%) (0%)
(3%) (5%) (88%) (4%) (1%) 0.22
70 (47%) 79 (53%)
222 (41%) 315 (59%) 0.21
13.40 1.82–37.89
13.74 0.83–44.84
Author’s calculations based on data from Independent Health insurance claims from 2010–2012. a Variables significantly related to Home Connections enrollment in the logistic regression model.
relatively equal at two years to six months prior to death. There were statistically significant differences from the last three months to the last two weeks of life, in which HC patients incurred lower costs, even with program fees included. At three months prior to death the average total allowed PMPM cost was $6,804 for those in HC, compared to $10,712 for those not enrolled. At one month prior to death the PMPM cost was $7,170 for HC versus $13,440 for controls. The trend continued with lower PMPM costs for HC members during the last two weeks of life as well, with total allowed PMPM cost $6,674 compared to $13,846. There was a highly significant effect of program membership over time within subjects in the repeated measures analysis of variance ( p-value < 0.0001). This effect remained highly significant in the multivariate analysis, which included adjustment for CHF and CRD ( p-value < 0.0001). The overall group effects between subjects were not statistically significant, however, when looking across the full two years
( p = 0.28). These results are consistent with the graphical comparison of program costs over time shown in Figure 1. The effects of program membership on the total overall medical and program costs during the last three months of life were quantified using a generalized linear model (see Table 3). There was a highly significant difference in costs between those enrolled in HC compared to controls, in which those enrolled in HC had significantly lower costs ( p-value < 0.0001). When adjusted for other clinical covariates, those in the HC program still had significantly lower costs ( p-value = 0.0002). Costs by type of service Facility. HC participants had significantly lower inpatient facility allowed costs than controls at all time points, with the greatest differences at the last one month and two weeks of life. Inpatient hospital admission rates were also much lower for HC participants, particularly during the last six months of life (see Fig. 2). Outpatient facility allowed
HOME-BASED PALLIATIVE CARE
1331
Table 2. Distribution of Medical Conditions between Home Connections Participants and Propensity-Matched Comparison Members
Condition Cancer diagnosisa Yes No CAD diagnosis Yes No CHF diagnosis Yes No Diabetes diagnosisa Yes No Depression diagnosisa Yes No COPD diagnosis Yes No CRD diagnosisa Yes No CVA diagnosis Yes No Dementia diagnosisa Yes No Alzheimer’s diagnosisa Yes No Asthma diagnosis Yes No
Home connections (n = 149) No. (%)
Comparison members (n = 537) No. (%)
87 (58%) 62 (42%)
280 (52%) 257 (48%)
55 (37%) 94 (63%)
218 (41%) 319 (59%)
42 (28%) 107 (72%)
213 (40%) 324 (60%)
39 (26%) 110 (74%)
155 (29%) 382 (71%)
30 (20%) 119 (80%)
78 (14%) 459 (85%)
27 (18%) 122 (82%)
117 (22%) 420 (78%)
21 (14%) 128 (86%)
150 (28%) 387 (72%)
20 (13%) 129 (87%)
103 (19%) 434 (81%)
15 (10%) 134 (90%)
68 (13%) 469 (87%)
11 (7%) 138 (93%)
48 (9%) 489 (91%)
8 (5%) 141 (95%)
21 (4%) 516 (96%)
p-value 0.19 0.45 0.01 0.54 0.10 0.36 < 0.0001 0.12 0.40 0.62 0.49
Author’s calculations based on data from Independent Health insurance claims from 2010–2012. a Variables significantly related to Home Connections enrollment in the logistic regression model.
costs were significantly higher among HC members at two years and one year. While no difference in outpatient costs was observed at six months, outpatient costs for HC members were significantly lower at three months, one month, and two weeks. There were no significant differences in emergency room allowed costs or utilization between HC members and controls at any time point analyzed. Pharmacy. Prescription costs were significantly higher for HC members across all time points except for one month and two weeks preceding death. Professional. Overall professional allowed costs were significantly higher among HC members at two years and one year. Conversely, while no difference in professional costs was observed at six months, costs for HC members were significantly lower at three months, one month, and two weeks. Further analysis of specialty care, a subset of professional costs, showed that specialist allowed costs were
significantly higher at one and two years for HC members, though utilization rates were essentially the same. The average allowed specialist PMPM cost and utilization were significantly lower among HC members in the last one month and two weeks of life. Ancillary. At two years, one year, and six months, HC members had significantly higher ancillary costs; however, there were no significant differences at three months, one month, or two weeks. There were highly significant differences in chemotherapy costs and utilization between HC members and the control group, in which HC members had both higher costs and utilization at one and two years. Fortyseven percent of HC members had received chemotherapy at one year, compared to only 25% of the control group. Similarly, at two years, half of HC members had received chemotherapy, compared to only 27% of controls. Chemotherapy costs and utilization were significantly higher for HC members at six months and three months preceding
1332
KERR ET AL.
FIG. 1. Comparison of overall total allowed PMPM costs over time and program costs between home connections and control group members. (Author’s calculations based on data from Independent Health insurance claims from 2010–2012; *indicates a statistically significant difference (p < 0.05.) death, but not significantly different from the control group at one month or two weeks. Radiation utilization was significantly higher for HC members at six months, three months, and one month prior to death. Significantly more HC participants—105 (70%) versus 133 (25%)—enrolled in hospice ( p < 0.0001). There was also a statistically significant difference in the Hospice.
length of time in hospice between HC members (median 34 days, range 0–276) and the control group (median 9 days, range 0–606) ( p = 0.0003). Discussion
Outpatient palliative care in the United States has been offered primarily under the Hospice Medicare Benefit, restricted
Table 3. Generalized Linear Model Analysis of Program Membership on Overall Total Costs during the Last Three Months of Life Variable Univariate model: Home Connections participation (yes vs no) Multivariate model: Home Connections participation (yes vs no) Age group (years) 0–50 51–65 66–75 76–90 ‡ 91 Cancer Dx (yes vs no) COPD Dx (yes vs no) Diabetes Dx (yes vs no) CHF Dx (yes vs no) CRD Dx (yes vs no) Prospective Risk Score
Parameter estimate
95% Confidence interval
p-value
- 0.42
- 0.62 to - 0.21
< 0.0001
- 0.37
- 0.56 to - 0.18
0.0002
0.17 0.41 0.39 Reference - 0.33 0.04 - 0.05 - 0.06 0.17 - 0.10 0.05
- 0.42 to 0.76 0.15 to 0.68 0.18 to 0.59 — - 0.60 to - 0.06 - 0.14 to 0.22 - 0.25 to 0.14 - 0.23 to 0.12 - 0.01 to 0.35 - 0.32 to 0.12 0.03 to 0.06
0.57 0.002 0.0003 — 0.02 0.65 0.59 0.50 0.07 0.36 < 0.0001
HOME-BASED PALLIATIVE CARE
1333
FIG. 2. Comparison of inpatient allowed PMPM costs over time between home connections and control member groups. (Author’s calculations based on data from Independent Health insurance claims from 2010–2012; *denotes statistically significant difference in costs for that time point.) to patients with expected prognoses of six months or less who are not pursuing curative treatments.3 While the number of inpatient palliative care programs has grown rapidly since 2000,4–9,22 there remains a strong need for innovative outpatient palliative care programs in order to provide services earlier in illness and throughout the care continuum.10 The current study presents the financial analysis of Home Connections (HC), a community-based outpatient palliative care program supported by Independent Health, a commercial insurance company, on end-of-life care. The program was cost effective over the study period, with costs the same or lower for HC members, even when program fees were included with patient medical expenditures. HC participants were nearly three times as likely to enroll in hospice near end of life, with a significantly longer length of stay. As reported separately,13 the program demonstrated significant clinical benefits for participants, including a high rate of actionable advance directive completion, improvement in multiple symptom domains, and an increased likelihood of dying at home. Patient, caregiver, and clinician satisfaction scores were very positive. Collectively, these findings demonstrate the significant potential of a home-based palliative care program to achieve the Institute of Healthcare Improvement’s Triple Aims23 goals of lower costs, higher quality, and improved satisfaction. Savings to commercial insurer
The most significant savings were apparent in the last three months of life, with lower overall, inpatient, outpatient, and professional costs. Although outpatient allowed costs and professional costs were higher among HC members at two years and one year, these costs fell significantly compared to
the control group during the last three months of life. Ancillary costs were also higher for HC participants at two years and one year, but were no longer significantly different during the last year of life. Hospital admissions, specialty visits, and their related costs for HC members were significantly lower than in the control group, and emergency costs showed no difference; while HC patients had significantly higher chemotherapy utilization until the three months preceding death. Although higher chemotherapy costs may be due to patient selection, further analysis of cancer types may help to understand the impact of the program at the service level. It is also conceivable that individuals who are not undergoing chemotherapy are more likely to be open to hospice referrals and thus already be in hospice care instead of a palliative care program. Examination of emergency room visit type may also help provide additional insights into these results. The types of patients referred to HC may have influenced higher costs earlier in the course of illness. With patient referrals to HC initiated by insurance company case managers and community physicians, it is likely that there was a selection bias towards referral of more complex ‘‘high cost, high utilizing’’ patients. Despite the likelihood that HC patients had greater medical complexity and utilization of higher-cost services, overall costs were equal at one and two years prior to death and lower at all later time points. Avoidance of hospitalizations likely contributed to cost neutrality at earlier time points and cost savings as patients approached end of life. The study was restricted to members who had expired, in order to eliminate bias introduced by deriving an equivalent enrollment date for the control group; further analysis is needed to understand the impact of the program on members who didn’t expire to see if cost benefits
1334
are similar. It is also unclear how well the results generalize to patients and health care systems in other settings. Enhanced hospice referral
Patients with serious illness often face highly fragmented care which may act as a barrier to hospice enrollment.24,25 In this study, 70% of patients entered hospice care, compared to 25% in the control group. Additionally, HC patients had a median length of stay in hospice that was nearly a month longer (34 versus 9 days). With the national median length of stay for hospice patients only 19 days,24 timelier referral to hospice, where the Hospice Medicare Benefit pays for all costs referable to the terminal illness, is an obvious cost savings to the commercial insurance company. A recent study showed significant savings continuing after hospice enrollment, with an estimated additional $6,430 saved per patient during a 15 to 30 day enrollment.25 In addition, hospice care has demonstrated clinical benefits including reduced hospitalizations, ED and ICU use, and reduced likelihood of death in-hospital for patients with cancer.27 In addition to potential financial benefits, patients dying at home experience greater quality of life, physical comfort, and psychological well-being, while family members and caregivers exhibit lower rates of pre-loss grief, posttraumatic stress disorder, anxiety, and depression.28 There are many barriers to hospice referral, including a reluctance to discuss end-of-life issues, lack of provider time, misconceptions surrounding the benefit, and the perceived need to ‘‘abandon’’ curative care.24 Interpretation of the benefit may also vary among hospice providers.29 Hospice referrals in HC were enhanced by the involvement of nurse case managers who were, for the most part, hospice trained RNs comfortable discussing end-of-life issues and helping patients and families define goals of care. Limitations
The use of propensity-score matching instead of a randomized control trial design may fail to adequately account for unmeasured variables that could explain some group differences, such as differences in program enrollee’s behavior or illness. Further, while many characteristics were evenly distributed between HC and control groups, there was a statistically significant difference in the incidence of CHF and CRD. Dissimilar distributions observed in the CHF and CRD populations may be a result of outreach efforts focusing on primary care and oncology practices. Also, as CRD patients on dialysis are traditionally underrepresented in hospice care, it appears that they were underrepresented in this palliative care program as well. This imbalance was adjusted for in the multivariate total cost analysis during the last three months of life. Additionally, although other diagnoses were distributed relatively evenly between HC and control groups, there was no comparison of severity of illness or comorbidities, beyond what is accounted for in the prospective risk score, an industry-acceptable measure of future cost, as determined by morbidity. Selection bias towards referral of medically complex patients to the HC program may have contributed to variations such as chemotherapy and radiation costs and utilization observed between HC enrollees and the control group. Patients were not randomly assigned to HC versus usual care,
KERR ET AL.
but were referred by insurance company case managers as well as community physicians based on a subjective perception of need and benefit. Abstraction of date of death from secondary sources may also have led to misclassification of some patients, although members with obvious discrepancies were excluded. Conclusions
Until recently, the Hospice Medicare Benefit has been the primary means to access home-based palliative care, restricted to the subset of patients no longer receiving curative or aggressive treatment with an expected prognosis of six months or less. Organizations who administer the hospice benefit typically are well positioned and expert in providing home-based palliative care; however, sustainable reimbursement models outside the Hospice Benefit have been absent or inadequate. As pointed out by Kelley, et al.,24 palliative care programs should develop formal relationships with community hospice providers to ‘‘offer a bridge to timely hospice enrollment,’’ thus realizing cost savings across the care continuum and avoiding unintended further care fragmentation. This study provides evidence that a commercial insurance company can realize significant cost savings while financially supporting a home-based palliative care program administered by a community hospice provider. Acknowledgments
Kathleen A. Donohue, MS and Sarah M. Knodel, MS, MPH had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Author Disclosure Statement
The authors have no conflicts of interest to report. This article shares a copyright with Independent Health, Williamsville, New York. References
1. U.S. Department of Health and Human Services, Centers for Medicare & Medicaid Services: Medicare Hospice Benefits. U.S. Department of Health and Human Services, Centers for Medicare & Medicaid Services. Baltimore, MD, 2013. 2. NIHCM Foundation: The Concentration of Health Care Spending. NIHCM Foundation, Washington, DC, 2012. 3. Center to Advance Palliative Care: Growth of Palliative Care in U.S. Hospitals 2012 Snapshot. Center to Advance Palliative Care, New York, NY, 2012. 4. Ciemins EL, Blum L, Nunley M, et al.: The economic and clinical impact of an inpatient palliative care consultation service: A multifaceted approach. J Palliat Med 2007;10: 1347–1355. 5. Gade G, Venohr I, Conner D, et al.: Impact of an inpatient palliative care team: A randomized control trial. J Palliat Med 2008;11:180–190. 6. Hanson LC, Usher B, Spragens L, Bernard S: Clinical and economic impact of palliative care consultation. J Pain Symptom Manage 2008;35:340–346. 7. Morrison RS, Dietrich J, Ladwig S, et al.: Palliative care consultation teams cut hospital costs for Medicaid beneficiaries. Health Aff (Millwood) 2011;30:454–463.
HOME-BASED PALLIATIVE CARE
8. Morrison RS, Maroney-Galin C, Kralovec PD, Meier DE: The growth of palliative care programs in United States hospitals. J Palliat Med 2005;8:1127–1134. 9. Morrison RS, Penrod JD, Cassel JB, et al.: Cost savings associated with US hospital palliative care consultation programs. Arch Intern Med 2008;168:1783–1790. 10. Rabow M, Kvale E, Barbour L, et al.: Moving upstream: A review of the evidence of the impact of outpatient palliative care. J Palliat Med 2013;12:1540–1549. 11. Brumley R, Enguidanos S, Jamison P, et al.: Increased satisfaction with care and lower costs: Results of a randomized trial of in-home palliative care. J Am Geriatr Soc 2007;55:993–1000. 12. Spettell CM, Rawlins WS, Krakauer R, et al.: A comprehensive case management program to improve palliative care. J Palliat Med 2009;12:827–832. 13. Kerr CW, Tangeman JC, Rudra CB, et al.: Clinical impacts of a home-based palliative care program: A hospice-private payer partnership. J Pain Symptom Manage 2014. 14. U.S. Census Bureau: Census 2000 Summary File 3, 2011. U.S. Census Bureau, 2011. 15. Murtagh FE, Preston M, Higginson I: Patterns of dying: Palliative care for non-malignant disease. Clin Med J 2004; 4:39–44. 16. Traue DC, Ross JR: Palliative care in non-malignant diseases. J R Soc Med 2005;98:503–506. 17. D’Agostino RB, Jr.: Propensity score methods for bias reduction in the comparison of a treatment to a nonrandomized control group. Stat Med 1998;17:2265–2281. 18. Bergstralh EJ, Kosanke JL, GMATCH: Computerized matching of cases to controls using the greedy matching algorithm with a fixed number of controls per case. Section of Biostatistics, Technical Report 56(1995). Mayo Clinic, Department of Biomedical Statistics and Informatics, Rochester, MN. 19. Cole JWL, and Grizzle JE: Applications of multivariate analysis of variance to repeated measures experiments. Biometrics 1966;22:810–828.
1335
20. Barber J, Thompson S: Multiple regression of cost data: Use of generalised linear models. J Health Serv Res Policy 2004;9:197–204. 21. SAS Institute: SAS/STAT 9.22 User’s Guide. Cary, NC: SAS Institute Inc; 2010. 22. Penrod JD, Deb P, Dellenbaugh C, et al.: Hospital-based palliative care consultation: Effects on hospital cost. J Palliat Med 2010;13:973–979. 23. Institute for Healthcare Improvement: IHI Triple Aim Initiative. http://www.ihi.org/offerings/Initiatives/TripleAim/ Pages/default.aspx. Accessed January 29, 2014. 24. The Hastings Center: Access to Hospice Care: Expanding Boundaries, Overcoming Barriers. Hastings, NY: The Hastings Center, 2003. 25. Kelley AS, Deb P, Du Q, et al.: Hospice enrollment saves money for Medicare and improves care quality across a number of different lengths-of-stay. Health Aff (Millwood) 2013;32:552–561. 26. NHPCO Facts and Figures: Hospice Care in America. National Hospice and Palliative Care Organization, Alexandria, VA, 2012. 27. Carlson MD, Herrin J, Du Q, et al.: Impact of hospice disenrollment on health care use and medicare expenditures for patients with cancer. J Clin Oncol 2010;28:4371–4375. 28. Wright AA, Keating NL, Balboni TA, et al.: Place of death: Correlations with quality of life of patients with cancer and predictors of bereaved caregivers’ mental health. J Clin Oncol 2010;28:4457–4464. 29. Carlson MD, Barry CL, Cherlin EJ, et al.: Hospices’ enrollment policies may contribute to underuse of hospice care in the United States. Health Aff (Millwood) 2012;31:2690–2698.
Address correspondence to: Pei C. Grant, PhD Center for Hospice and Palliative Care 225 Como Park Boulevard Cheektowaga, NY 14227 E-mail: [email protected]
Cost Savings and Enhanced Hospice Enrollment with a Home-Based Palliative Care Program Implemented as a Hospice–Private Payer Partnership Christopher W. Kerr, MD, PhD,1 Kathleen A. Donohue, MA,2 John C. Tangeman, MD, FACP,1 Amin M. Serehali, MA,2 Sarah M. Knodel, MS, MPH,2 Pei C. Grant, PhD,1 Debra L. Luczkiewicz, MD,1 Kathleen Mylotte, MD,2 and Melanie J. Marien, MS, PA-C1
Abstract
Background: In the United States, 5% of the population is responsible for nearly half of all health care expenditures, with a large concentration of spending driven by individuals with expensive chronic conditions in their last year of life. Outpatient palliative care under the Medicare Hospice Benefit excludes a large proportion of the chronically ill and there is widespread recognition that innovative strategies must be developed to meet the needs of the seriously ill while reducing costs. Objective: This study aimed to evaluate the impact of a home-based palliative care program, implemented through a hospice–private payer partnership, on health care costs and utilization. Methods: This was a prospective, observational database study where insurance enrollment and claims data were analyzed. The study population consisted of Home Connections (HC) program patients enrolled between January 1, 2010 and December 31, 2012 who subsequently expired (n = 149) and who were also Independent Health members. A control group (n = 537) was derived using propensity-score matching. The primary outcome variable was overall costs within the last year of life. Costs were also examined at six months, three months, one month, and two weeks. Inpatient, outpatient, ancillary, professional, and pharmacy costs were compared between the two groups. Medical service utilization and hospice enrollment and length of stay were also evaluated. Results: Cost savings were apparent in the last three months of life—$6,804 per member per month (PMPM) cost for palliative care participants versus $10,712 for usual care. During the last two weeks of life, total allowed PMPM was $6,674 versus $13,846 for usual care. Enhanced hospice entry (70% versus 25%) and longer length of stay in hospice (median 34 versus 9 days) were observed. Conclusions: Palliative care programs partnered with community hospice providers may achieve cost savings while helping provide care across the continuum. Introduction
O
utpatient palliative care in most communities has been administered primarily via the Hospice Medicare Benefit which provides care for terminally ill individuals with an estimated life expectancy of six months or less. In addition, patients under the Hospice Medicare Benefit have chosen to ‘‘forgo curative treatment,’’1 which may exclude many patients facing chronic, incurable illness who might otherwise benefit from palliative care. Health care systems and payers are seeking innovative programs to address the needs of the most seriously ill patients who incur the highest
costs; in 2009, the sickest 5% of patients in the United States accounted for greater than 50% of health care spending, with a large proportion spent in the last year of life, often on hospital and/or ICU care.2 Hospital-based palliative care consultation services have, over the past decade, been offered at an increasing number of large U.S. hospitals,3 with a solid evidence base supporting financial and clinical efficacy.4–9 As noted in a recent review, less is known about the impact of outpatient palliative care programs.10 Palliative care offers the rare alignment between positive clinical and economic outcomes. A previous study of homebased palliative care showed reduced costs of up to 33%;11
1
Center for Hospice and Palliative Care, Cheektowaga, New York. Independent Health, Buffalo, New York. Accepted August 14, 2014. This article shares copyright with Independent Health, Williamsville, New York.
2
1328
HOME-BASED PALLIATIVE CARE
however, this program was implemented through a staff model insurance company and may be difficult to replicate outside of this framework. Similarly, the Aetna Compassionate Care Program showed a 22% reduction in costs at end of life while doubling hospice utilization.12 The current study describes economic outcomes of a collaborative model between a notfor-profit hospice and a private not-for-profit commercial insurance payer to provide home-based palliative care upstream and outside of hospice. This innovative program, implemented in Erie County in western New York, was supported via a per member per month (PMPM) fee. In a previous study,13 the clinical impact of this program was evaluated and was shown to improve symptom management in multiple symptom domains. More than two thirds of enrollees completed actionable advance directives, and the site of death was home for nearly half of those who expired during or after participation. Additionally, patients, caregivers, and physicians all gave the program high satisfaction scores.13 An economic analysis of participants in the program who expired is now reported, including the impact of this model on overall end-of-life costs, service-specific costs, and medical service utilization. Hospice enrollment and length of stay are also discussed. Methods Program description
The Center for Hospice & Palliative Care, Cheektowaga, New York, established the Home Connections (HC) homebased palliative care program in 2008 jointly with Independent Health, a commercial insurance provider. HC is intended for adults with advanced chronic illness upstream of hospice, so patients may still be receiving aggressive or curefocused treatments. The program is not restricted to patients with a prognosis of six months or less. The HC team includes a palliative care-trained registered nurse (RN), social worker, volunteers, and palliative care physician. Services include symptom management, education, supportive discussions about heath care decision making and goals, social work visits to facilitate access to community support services, respite care, and 24/7 on-call palliative care nurse support. Nurses and social workers visit participants a minimum of every 30 days, and a biweekly interdisciplinary group meeting is held to discuss each patient case with the palliative care physician. When clinically appropriate, HC also helps patients transition into hospice care, if desired. Sample
This study was approved by the University at Buffalo Health Sciences IRB. The study population consisted of HC participants enrolled for greater than 30 days between January 1, 2010 and December 31, 2012 who subsequently expired (n = 149). A control group was derived using propensity-score matching, to compare members with similar clinical and demographic characteristics (n = 537). Members enrolled less than 30 days (n = 68) were excluded, as many of those members enrolled directly in hospice (63%), therefore limiting the ability to measure the effect of HC on cost and utilization over a defined time period. Data was abstracted from insurance enrollment and claims. Logistic regression analysis was used to identify control group patients for propensity matching using age, gender,
1329
insurance product, median household income (categorized), neighborhood percentage minority (categorized), prospective risk score,14 and diagnoses historically associated with palliative care,5,16 These diagnoses included cancer, asthma, coronary artery disease (CAD), chronic obstructive pulmonary disease (COPD), diabetes, congestive heart failure (CHF), cerebral vascular accident (CVA), Alzheimer’s disease, dementia, chronic renal disease (CRD), and depression. The predicted probability of HC membership was grouped into quintiles, following the methodology of R. D’Agostino.17 The SAS GMATCH algorithm18 matched HC members to controls, at a 1:4 ratio. Variables found to be significantly related to participation in HC included age; neighborhood median household income, neighborhood percent minority race; diagnoses of cancer, diabetes, Alzheimer’s disease, dementia, CRD, and depression; insurance product; and prospective risk score. Analysis was restricted to members who had expired in order to eliminate bias introduced by trying to ‘‘match’’ clinical characteristics based on dates of enrollment. The data were filtered to only include those in the range of probability scores for HC members, yielding 5400 members in the pool used to match cases to controls. Statistical analyses
Outcome measures were costs and utilization of services, evaluated at two years, one year, six months, three months, one month, and two weeks prior to death. Testing of categorical factors was performed using the chi-square or Fisher’s exact test. Cost and risk score data were analyzed using the MannWhitney test. Statistical significance was set at 5% (two-sided). The effects of program membership on overall costs were analyzed using a repeated measures analysis of variance.19 This model allowed for changes in cost to be measured between the members based on program membership as well as over time. Cost data in the repeated measures analysis was transformed using a log function to adjust for skewing and outliers. The cost total within the last three months of life was modeled as a gamma distribution using a log link function20 using the generalized linear model procedure.21 Multivariate analysis adjusted for the effects of other demographic and clinical characteristics on overall costs. Results HC participants and comparison members
The median time in HC, excluding those enrolled < 30 days, was 118 days (range: 31–843). Demographic characteristics of HC participants and propensity-matched comparison members are shown in Table 1, with medical diagnoses in Table 2. Propensity score quintiles were evenly distributed, indicating that the groups were well matched for the likelihood of participation in HC. Demographic characteristics and diagnosis were similarly distributed between HC members and controls with the exception of CHF (40% of controls versus 28% of HC members [p = 0.01]) and CRD (28% of controls versus 14% of HC participants [p < 0.0001]). Aggregate costs
A comparison of overall total allowed costs over time is shown in Figure 1. Costs between the two groups were
1330
KERR ET AL.
Table 1. Distribution of Demographic Characteristics between Home Connections Participants and Propensity-Matched Comparison Members
Characteristic Propensity score quintile 1 2 3 4 5 Age group (years)a 0–50 51–65 66–75 76–90 ‡ 91 Neighborhood median household income ($)a £ 10,000 10,001–20,000 20,001–30,000 30,001–50,000 50,001–75,000 ‡ 75,001 Neighborhood percent minority racea < 50% ‡ 50% Insurance producta Article 43 Encompass Medicare Medisource Self-funded Gender Male Female Prospective risk scorea Median Range
Home connections (n = 149) No. (%)
Comparison members (n = 537) No. (%)
p-value 0.99
29 29 32 29 30
(19%) (19%) (21%) (19%) (20%)
107 113 108 104 105
(20%) (21%) (20%) (19%) (20%)
4 20 36 80 9
(3%) (13%) (24%) (53%) (6%)
9 71 116 280 61
(2%) (13%) (22%) (52%) (11%)
2 1 12 60 45 29
(1%) ( < 1%) (8%) (40%) (30%) (19%)
6 9 52 183 208 79
(1%) (2%) (10%) (34%) (39%) (15%)
0.37
0.26
0.60 136 (91%) 13 (9%)
498 (93%) 39 (7%)
5 10 131 3 0
15 27 470 19 6
0.52 (3%) (7%) (88%) (2%) (0%)
(3%) (5%) (88%) (4%) (1%) 0.22
70 (47%) 79 (53%)
222 (41%) 315 (59%) 0.21
13.40 1.82–37.89
13.74 0.83–44.84
Author’s calculations based on data from Independent Health insurance claims from 2010–2012. a Variables significantly related to Home Connections enrollment in the logistic regression model.
relatively equal at two years to six months prior to death. There were statistically significant differences from the last three months to the last two weeks of life, in which HC patients incurred lower costs, even with program fees included. At three months prior to death the average total allowed PMPM cost was $6,804 for those in HC, compared to $10,712 for those not enrolled. At one month prior to death the PMPM cost was $7,170 for HC versus $13,440 for controls. The trend continued with lower PMPM costs for HC members during the last two weeks of life as well, with total allowed PMPM cost $6,674 compared to $13,846. There was a highly significant effect of program membership over time within subjects in the repeated measures analysis of variance ( p-value < 0.0001). This effect remained highly significant in the multivariate analysis, which included adjustment for CHF and CRD ( p-value < 0.0001). The overall group effects between subjects were not statistically significant, however, when looking across the full two years
( p = 0.28). These results are consistent with the graphical comparison of program costs over time shown in Figure 1. The effects of program membership on the total overall medical and program costs during the last three months of life were quantified using a generalized linear model (see Table 3). There was a highly significant difference in costs between those enrolled in HC compared to controls, in which those enrolled in HC had significantly lower costs ( p-value < 0.0001). When adjusted for other clinical covariates, those in the HC program still had significantly lower costs ( p-value = 0.0002). Costs by type of service Facility. HC participants had significantly lower inpatient facility allowed costs than controls at all time points, with the greatest differences at the last one month and two weeks of life. Inpatient hospital admission rates were also much lower for HC participants, particularly during the last six months of life (see Fig. 2). Outpatient facility allowed
HOME-BASED PALLIATIVE CARE
1331
Table 2. Distribution of Medical Conditions between Home Connections Participants and Propensity-Matched Comparison Members
Condition Cancer diagnosisa Yes No CAD diagnosis Yes No CHF diagnosis Yes No Diabetes diagnosisa Yes No Depression diagnosisa Yes No COPD diagnosis Yes No CRD diagnosisa Yes No CVA diagnosis Yes No Dementia diagnosisa Yes No Alzheimer’s diagnosisa Yes No Asthma diagnosis Yes No
Home connections (n = 149) No. (%)
Comparison members (n = 537) No. (%)
87 (58%) 62 (42%)
280 (52%) 257 (48%)
55 (37%) 94 (63%)
218 (41%) 319 (59%)
42 (28%) 107 (72%)
213 (40%) 324 (60%)
39 (26%) 110 (74%)
155 (29%) 382 (71%)
30 (20%) 119 (80%)
78 (14%) 459 (85%)
27 (18%) 122 (82%)
117 (22%) 420 (78%)
21 (14%) 128 (86%)
150 (28%) 387 (72%)
20 (13%) 129 (87%)
103 (19%) 434 (81%)
15 (10%) 134 (90%)
68 (13%) 469 (87%)
11 (7%) 138 (93%)
48 (9%) 489 (91%)
8 (5%) 141 (95%)
21 (4%) 516 (96%)
p-value 0.19 0.45 0.01 0.54 0.10 0.36 < 0.0001 0.12 0.40 0.62 0.49
Author’s calculations based on data from Independent Health insurance claims from 2010–2012. a Variables significantly related to Home Connections enrollment in the logistic regression model.
costs were significantly higher among HC members at two years and one year. While no difference in outpatient costs was observed at six months, outpatient costs for HC members were significantly lower at three months, one month, and two weeks. There were no significant differences in emergency room allowed costs or utilization between HC members and controls at any time point analyzed. Pharmacy. Prescription costs were significantly higher for HC members across all time points except for one month and two weeks preceding death. Professional. Overall professional allowed costs were significantly higher among HC members at two years and one year. Conversely, while no difference in professional costs was observed at six months, costs for HC members were significantly lower at three months, one month, and two weeks. Further analysis of specialty care, a subset of professional costs, showed that specialist allowed costs were
significantly higher at one and two years for HC members, though utilization rates were essentially the same. The average allowed specialist PMPM cost and utilization were significantly lower among HC members in the last one month and two weeks of life. Ancillary. At two years, one year, and six months, HC members had significantly higher ancillary costs; however, there were no significant differences at three months, one month, or two weeks. There were highly significant differences in chemotherapy costs and utilization between HC members and the control group, in which HC members had both higher costs and utilization at one and two years. Fortyseven percent of HC members had received chemotherapy at one year, compared to only 25% of the control group. Similarly, at two years, half of HC members had received chemotherapy, compared to only 27% of controls. Chemotherapy costs and utilization were significantly higher for HC members at six months and three months preceding
1332
KERR ET AL.
FIG. 1. Comparison of overall total allowed PMPM costs over time and program costs between home connections and control group members. (Author’s calculations based on data from Independent Health insurance claims from 2010–2012; *indicates a statistically significant difference (p < 0.05.) death, but not significantly different from the control group at one month or two weeks. Radiation utilization was significantly higher for HC members at six months, three months, and one month prior to death. Significantly more HC participants—105 (70%) versus 133 (25%)—enrolled in hospice ( p < 0.0001). There was also a statistically significant difference in the Hospice.
length of time in hospice between HC members (median 34 days, range 0–276) and the control group (median 9 days, range 0–606) ( p = 0.0003). Discussion
Outpatient palliative care in the United States has been offered primarily under the Hospice Medicare Benefit, restricted
Table 3. Generalized Linear Model Analysis of Program Membership on Overall Total Costs during the Last Three Months of Life Variable Univariate model: Home Connections participation (yes vs no) Multivariate model: Home Connections participation (yes vs no) Age group (years) 0–50 51–65 66–75 76–90 ‡ 91 Cancer Dx (yes vs no) COPD Dx (yes vs no) Diabetes Dx (yes vs no) CHF Dx (yes vs no) CRD Dx (yes vs no) Prospective Risk Score
Parameter estimate
95% Confidence interval
p-value
- 0.42
- 0.62 to - 0.21
< 0.0001
- 0.37
- 0.56 to - 0.18
0.0002
0.17 0.41 0.39 Reference - 0.33 0.04 - 0.05 - 0.06 0.17 - 0.10 0.05
- 0.42 to 0.76 0.15 to 0.68 0.18 to 0.59 — - 0.60 to - 0.06 - 0.14 to 0.22 - 0.25 to 0.14 - 0.23 to 0.12 - 0.01 to 0.35 - 0.32 to 0.12 0.03 to 0.06
0.57 0.002 0.0003 — 0.02 0.65 0.59 0.50 0.07 0.36 < 0.0001
HOME-BASED PALLIATIVE CARE
1333
FIG. 2. Comparison of inpatient allowed PMPM costs over time between home connections and control member groups. (Author’s calculations based on data from Independent Health insurance claims from 2010–2012; *denotes statistically significant difference in costs for that time point.) to patients with expected prognoses of six months or less who are not pursuing curative treatments.3 While the number of inpatient palliative care programs has grown rapidly since 2000,4–9,22 there remains a strong need for innovative outpatient palliative care programs in order to provide services earlier in illness and throughout the care continuum.10 The current study presents the financial analysis of Home Connections (HC), a community-based outpatient palliative care program supported by Independent Health, a commercial insurance company, on end-of-life care. The program was cost effective over the study period, with costs the same or lower for HC members, even when program fees were included with patient medical expenditures. HC participants were nearly three times as likely to enroll in hospice near end of life, with a significantly longer length of stay. As reported separately,13 the program demonstrated significant clinical benefits for participants, including a high rate of actionable advance directive completion, improvement in multiple symptom domains, and an increased likelihood of dying at home. Patient, caregiver, and clinician satisfaction scores were very positive. Collectively, these findings demonstrate the significant potential of a home-based palliative care program to achieve the Institute of Healthcare Improvement’s Triple Aims23 goals of lower costs, higher quality, and improved satisfaction. Savings to commercial insurer
The most significant savings were apparent in the last three months of life, with lower overall, inpatient, outpatient, and professional costs. Although outpatient allowed costs and professional costs were higher among HC members at two years and one year, these costs fell significantly compared to
the control group during the last three months of life. Ancillary costs were also higher for HC participants at two years and one year, but were no longer significantly different during the last year of life. Hospital admissions, specialty visits, and their related costs for HC members were significantly lower than in the control group, and emergency costs showed no difference; while HC patients had significantly higher chemotherapy utilization until the three months preceding death. Although higher chemotherapy costs may be due to patient selection, further analysis of cancer types may help to understand the impact of the program at the service level. It is also conceivable that individuals who are not undergoing chemotherapy are more likely to be open to hospice referrals and thus already be in hospice care instead of a palliative care program. Examination of emergency room visit type may also help provide additional insights into these results. The types of patients referred to HC may have influenced higher costs earlier in the course of illness. With patient referrals to HC initiated by insurance company case managers and community physicians, it is likely that there was a selection bias towards referral of more complex ‘‘high cost, high utilizing’’ patients. Despite the likelihood that HC patients had greater medical complexity and utilization of higher-cost services, overall costs were equal at one and two years prior to death and lower at all later time points. Avoidance of hospitalizations likely contributed to cost neutrality at earlier time points and cost savings as patients approached end of life. The study was restricted to members who had expired, in order to eliminate bias introduced by deriving an equivalent enrollment date for the control group; further analysis is needed to understand the impact of the program on members who didn’t expire to see if cost benefits
1334
are similar. It is also unclear how well the results generalize to patients and health care systems in other settings. Enhanced hospice referral
Patients with serious illness often face highly fragmented care which may act as a barrier to hospice enrollment.24,25 In this study, 70% of patients entered hospice care, compared to 25% in the control group. Additionally, HC patients had a median length of stay in hospice that was nearly a month longer (34 versus 9 days). With the national median length of stay for hospice patients only 19 days,24 timelier referral to hospice, where the Hospice Medicare Benefit pays for all costs referable to the terminal illness, is an obvious cost savings to the commercial insurance company. A recent study showed significant savings continuing after hospice enrollment, with an estimated additional $6,430 saved per patient during a 15 to 30 day enrollment.25 In addition, hospice care has demonstrated clinical benefits including reduced hospitalizations, ED and ICU use, and reduced likelihood of death in-hospital for patients with cancer.27 In addition to potential financial benefits, patients dying at home experience greater quality of life, physical comfort, and psychological well-being, while family members and caregivers exhibit lower rates of pre-loss grief, posttraumatic stress disorder, anxiety, and depression.28 There are many barriers to hospice referral, including a reluctance to discuss end-of-life issues, lack of provider time, misconceptions surrounding the benefit, and the perceived need to ‘‘abandon’’ curative care.24 Interpretation of the benefit may also vary among hospice providers.29 Hospice referrals in HC were enhanced by the involvement of nurse case managers who were, for the most part, hospice trained RNs comfortable discussing end-of-life issues and helping patients and families define goals of care. Limitations
The use of propensity-score matching instead of a randomized control trial design may fail to adequately account for unmeasured variables that could explain some group differences, such as differences in program enrollee’s behavior or illness. Further, while many characteristics were evenly distributed between HC and control groups, there was a statistically significant difference in the incidence of CHF and CRD. Dissimilar distributions observed in the CHF and CRD populations may be a result of outreach efforts focusing on primary care and oncology practices. Also, as CRD patients on dialysis are traditionally underrepresented in hospice care, it appears that they were underrepresented in this palliative care program as well. This imbalance was adjusted for in the multivariate total cost analysis during the last three months of life. Additionally, although other diagnoses were distributed relatively evenly between HC and control groups, there was no comparison of severity of illness or comorbidities, beyond what is accounted for in the prospective risk score, an industry-acceptable measure of future cost, as determined by morbidity. Selection bias towards referral of medically complex patients to the HC program may have contributed to variations such as chemotherapy and radiation costs and utilization observed between HC enrollees and the control group. Patients were not randomly assigned to HC versus usual care,
KERR ET AL.
but were referred by insurance company case managers as well as community physicians based on a subjective perception of need and benefit. Abstraction of date of death from secondary sources may also have led to misclassification of some patients, although members with obvious discrepancies were excluded. Conclusions
Until recently, the Hospice Medicare Benefit has been the primary means to access home-based palliative care, restricted to the subset of patients no longer receiving curative or aggressive treatment with an expected prognosis of six months or less. Organizations who administer the hospice benefit typically are well positioned and expert in providing home-based palliative care; however, sustainable reimbursement models outside the Hospice Benefit have been absent or inadequate. As pointed out by Kelley, et al.,24 palliative care programs should develop formal relationships with community hospice providers to ‘‘offer a bridge to timely hospice enrollment,’’ thus realizing cost savings across the care continuum and avoiding unintended further care fragmentation. This study provides evidence that a commercial insurance company can realize significant cost savings while financially supporting a home-based palliative care program administered by a community hospice provider. Acknowledgments
Kathleen A. Donohue, MS and Sarah M. Knodel, MS, MPH had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Author Disclosure Statement
The authors have no conflicts of interest to report. This article shares a copyright with Independent Health, Williamsville, New York. References
1. U.S. Department of Health and Human Services, Centers for Medicare & Medicaid Services: Medicare Hospice Benefits. U.S. Department of Health and Human Services, Centers for Medicare & Medicaid Services. Baltimore, MD, 2013. 2. NIHCM Foundation: The Concentration of Health Care Spending. NIHCM Foundation, Washington, DC, 2012. 3. Center to Advance Palliative Care: Growth of Palliative Care in U.S. Hospitals 2012 Snapshot. Center to Advance Palliative Care, New York, NY, 2012. 4. Ciemins EL, Blum L, Nunley M, et al.: The economic and clinical impact of an inpatient palliative care consultation service: A multifaceted approach. J Palliat Med 2007;10: 1347–1355. 5. Gade G, Venohr I, Conner D, et al.: Impact of an inpatient palliative care team: A randomized control trial. J Palliat Med 2008;11:180–190. 6. Hanson LC, Usher B, Spragens L, Bernard S: Clinical and economic impact of palliative care consultation. J Pain Symptom Manage 2008;35:340–346. 7. Morrison RS, Dietrich J, Ladwig S, et al.: Palliative care consultation teams cut hospital costs for Medicaid beneficiaries. Health Aff (Millwood) 2011;30:454–463.
HOME-BASED PALLIATIVE CARE
8. Morrison RS, Maroney-Galin C, Kralovec PD, Meier DE: The growth of palliative care programs in United States hospitals. J Palliat Med 2005;8:1127–1134. 9. Morrison RS, Penrod JD, Cassel JB, et al.: Cost savings associated with US hospital palliative care consultation programs. Arch Intern Med 2008;168:1783–1790. 10. Rabow M, Kvale E, Barbour L, et al.: Moving upstream: A review of the evidence of the impact of outpatient palliative care. J Palliat Med 2013;12:1540–1549. 11. Brumley R, Enguidanos S, Jamison P, et al.: Increased satisfaction with care and lower costs: Results of a randomized trial of in-home palliative care. J Am Geriatr Soc 2007;55:993–1000. 12. Spettell CM, Rawlins WS, Krakauer R, et al.: A comprehensive case management program to improve palliative care. J Palliat Med 2009;12:827–832. 13. Kerr CW, Tangeman JC, Rudra CB, et al.: Clinical impacts of a home-based palliative care program: A hospice-private payer partnership. J Pain Symptom Manage 2014. 14. U.S. Census Bureau: Census 2000 Summary File 3, 2011. U.S. Census Bureau, 2011. 15. Murtagh FE, Preston M, Higginson I: Patterns of dying: Palliative care for non-malignant disease. Clin Med J 2004; 4:39–44. 16. Traue DC, Ross JR: Palliative care in non-malignant diseases. J R Soc Med 2005;98:503–506. 17. D’Agostino RB, Jr.: Propensity score methods for bias reduction in the comparison of a treatment to a nonrandomized control group. Stat Med 1998;17:2265–2281. 18. Bergstralh EJ, Kosanke JL, GMATCH: Computerized matching of cases to controls using the greedy matching algorithm with a fixed number of controls per case. Section of Biostatistics, Technical Report 56(1995). Mayo Clinic, Department of Biomedical Statistics and Informatics, Rochester, MN. 19. Cole JWL, and Grizzle JE: Applications of multivariate analysis of variance to repeated measures experiments. Biometrics 1966;22:810–828.
1335
20. Barber J, Thompson S: Multiple regression of cost data: Use of generalised linear models. J Health Serv Res Policy 2004;9:197–204. 21. SAS Institute: SAS/STAT 9.22 User’s Guide. Cary, NC: SAS Institute Inc; 2010. 22. Penrod JD, Deb P, Dellenbaugh C, et al.: Hospital-based palliative care consultation: Effects on hospital cost. J Palliat Med 2010;13:973–979. 23. Institute for Healthcare Improvement: IHI Triple Aim Initiative. http://www.ihi.org/offerings/Initiatives/TripleAim/ Pages/default.aspx. Accessed January 29, 2014. 24. The Hastings Center: Access to Hospice Care: Expanding Boundaries, Overcoming Barriers. Hastings, NY: The Hastings Center, 2003. 25. Kelley AS, Deb P, Du Q, et al.: Hospice enrollment saves money for Medicare and improves care quality across a number of different lengths-of-stay. Health Aff (Millwood) 2013;32:552–561. 26. NHPCO Facts and Figures: Hospice Care in America. National Hospice and Palliative Care Organization, Alexandria, VA, 2012. 27. Carlson MD, Herrin J, Du Q, et al.: Impact of hospice disenrollment on health care use and medicare expenditures for patients with cancer. J Clin Oncol 2010;28:4371–4375. 28. Wright AA, Keating NL, Balboni TA, et al.: Place of death: Correlations with quality of life of patients with cancer and predictors of bereaved caregivers’ mental health. J Clin Oncol 2010;28:4457–4464. 29. Carlson MD, Barry CL, Cherlin EJ, et al.: Hospices’ enrollment policies may contribute to underuse of hospice care in the United States. Health Aff (Millwood) 2012;31:2690–2698.
Address correspondence to: Pei C. Grant, PhD Center for Hospice and Palliative Care 225 Como Park Boulevard Cheektowaga, NY 14227 E-mail: [email protected]
